Abstract:
The goal of my master’s thesis is to find a theoretical explanation for mechanical
failure of SiC in power electronic devices. This is accomplished using first-principles
density functional theory calculations to determine the energetics of stacking faults
in SiC and related group IV semiconducting crystals (C, Si and Ge) for comparison.
SiC, unlike other group IV semiconductors, exhibits numerous polytypical forms
such as 3C, 4H and 6H, which differ only in the stacking sequence along the (111)
direction of its diamond (3C) structure.
First, we established that in all the above semiconductors, the relevant slip system
is always the on the glide plane. Secondly, the stacking fault energies of SiC
polytypes are much smaller than those of the elemental semiconductors. All these
estimates agree well with experiment, wherever the measured values are available.
The generalized stacking fault energy surfaces determined for the glide planes of
C, Si, Ge, 3C-SiC, 4H-SiC and 6H-SiC were used in statistical thermodynamical
analysis to get relative stability of the faulted crystal with respect to the perfect
crystal as a function of temperature. We demonstrated that SiC is quite different
from the elemental semiconductors: above a certain Tc, the faulted SiC crystal
becomes more stable than the perfect crystal, proving that there should be rapid
expansion of the stacking faults in SiC if the operating temperatures in high power
devices exceed Tc.
To identify the fundamental difference between SiC and other semiconductors, we
used the Axial Next Nearest Neighbour Model (ANNNI) to map the stacking fault
energies of these semiconductor crystals. We find that the ANNNI model captures
the energetics of all these semiconductors accurately. The nearest neighbour
parameter in this model is found to be significantly smaller in SiC than in other
systems.
Using modern crystal growth techniques it is possible to grow various polytypes
of any semiconductor. The stacking fault energies in such atomically engineered
ix
structures can be predicted easily using ANNNI model. We present our theoretical
predictions of the stacking fault energies of such crystal structures. We also prove
that 6H- polytype of any semiconductor would always have very low stacking fault
energy. Our results are expected to be useful in estimating the plastic behaviour
of these novel polytypes.
In short, the thesis explains the phenomenon of stacking fault expansion seen in
hexagonal polytypes of SiC. Our work has resulted in ANNNI models for various
semiconductors that can be used to predict stacking fault energies in atomically
engineered semiconductor structures.